It is known that transmission of a digital signal through a medium, whether wired or wireless, can result in the digital signal becoming distorted due to affects of the medium. Digital signals transmitted though a wireless medium (i.e., wireless signals) can be distorted due to effects of multipath fading, magnitude response, and phase response of the wireless medium. Digital signals transmitted though a wire medium (i.e., wired signals) can also be distorted due to effects of magnitude response and phase response of the wired medium. Distortion of a digital signal can lead to loss of data, or data that is not properly interpreted by a receiver of the digital signal.
A variety of methods of signal coding have been developed to address the above issues, particularly for wireless media. To be successful, however, most of these methods require prior knowledge of the medium fading and attenuation characteristics. In static situations, where the medium (i.e., channel) does not change with respect to time (e.g., a wired channel or a static wireless channel), a pre-characterization, in which characteristics of the channel are measured before a transmission of data, is sufficient to characterize the channel. Channel fading and attenuation effects can be measured in the pre-characterization, and corrections can be used to compensate at a receiving system.
However, in many practical applications, channel characteristics vary with respect to time. For example, a wireless transmitter or a wireless receiver can be moving, resulting in a changing communication channel. In these cases, some form of dynamic channel characterization is required to characterize the channel as it changes over time.
A common approach for performing the dynamic channel characterization is to incorporate one or more reference or calibration signals into a digital signal being transmitted in a communication. The calibration signals are often referred to as pilot tones, or pilot symbols. The format and content of the pilot tones are predetermined, and therefore, known at the receiver. When the pilot tones are detected at the receiver, any one of various algorithms can be used in order to derive a desired channel compensation by a comparison of characteristics of the known pilot tones to characteristics of the received pilot tones. For example, the amplitude and a phase of the pilot tones can be used for comparison.
A disadvantage of using pilot tones is that the pilot tones carry no data. Therefore, use of the pilot tones results in a loss of channel bandwidth that could otherwise be used for data transmission. The loss of bandwidth is exacerbated if the channel changes rapidly, wherein the pilot tones must be transmitted more often in order to dynamically update the channel characteristics used by the receiver for compensation.
To be of general utility in a practical environment, wireless communications, and in particular, wireless digital communications, must be robust in a variety of static and dynamic applications having multipath fading, attenuation, and other losses that degrade a transmitted digital signal as it propagates through a wireless medium to a receiver. The office environment typifies a somewhat static environment. In the office, the office occupants move about, but the principal sources of multipath reflections, such as furniture and walls, are generally fixed in place. Therefore, a wireless communication channel in an office environment can have relatively constant channel characteristics.
In contrast, mobile users, such as pedestrians and motorists in an urban setting, represent dynamic situations in which the characteristics (i.e., magnitude and phase response) of a communication channel change greatly and sometimes rapidly. In the case of the motorist using, for example, a cellular telephone, changes to characteristics of the wireless channel (or medium) occur very rapidly in the presence of movement, which can correspond to movement of several wavelengths in as little as 10 ms. The communication channel used by the motorist needs frequent characterizations using a multiplicity of pilot symbols. For the pedestrian using, for example, a cellular telephone, changes to characteristics of the wireless medium occur less rapidly and characterization of the channel are needed less often.
A ship, and especially a military ship, represents a unique and particularly challenging environment in which a wireless signal must propagate through a communication channel comprised of Faraday cages, i.e., all-metal compartments. Furthermore, reflective surfaces within the compartments are subject to movement. For example, airplanes and equipment in the hanger deck of an aircraft carrier are frequently moved over distances much greater than a wavelength of the wireless signal. Helmets, vehicles, and landing craft in the loading area, i.e., the well deck, of an amphibious landing ship, also move about the ship. Within the challenging shipboard environment, channel characteristics are subject to dynamic change at rapid rates.
Numerous coding approaches have been developed to maintain signal quality in the presence of multipath fading, attenuation, and other losses. However, most of these methods require prior knowledge of channel characteristics to optimally reconstruct and decode the received signal. This can be an issue in situations where channel characteristics are changing rapidly.
Space-time coding, known to those of ordinary skill in the art, is an example of one of the many forms of signal coding techniques used to achieve diversity gain (e.g., multiple channels) to operate in the presence of multipath fading, attenuation, and other problems. However, most forms of space-time coding require knowledge of channel characteristics.
Block coding, also known to those of ordinary skill in the art, is a computationally straightforward method using space-time coding. Block coding also requires knowledge of channel characteristics.
In order to obtain the knowledge of channel characteristics, a method of characterization of a communication channel is needed. For static cases, such as the typical office environment, measurement of characteristics (e.g., frequency response and phase response) of the channel prior to use may be sufficient. For dynamic cases, where channel characteristics change with respect to time, frequent measurement of the channel characteristics is necessary.
As described above, pilot tones embedded in a digital data stream can be used to provide the dynamic channel characterization, but at the expense of channel bandwidth. Some examples of conventional pilot tones include:
pilot tones described in IEEE 802.11; and
pilot tones transmitted at the same time as the data, but at a different frequency, for example, pilot tones transmitted in the sub-channels of OFDM (orthogonal frequency division multiplexing).
Methods for measuring communication channel characteristics from transmitted pilot tones generally fall into three categories: correlation, Best Linear Unbiased Estimator (BLUE), or Minimum Mean Square Error (MMSE). Computational complexity increases from correlation to BLUE to MMSE.
Correlation provides a direct comparison of a received pilot tone signal to an expected pilot tone signal, using correlation techniques. The expected pilot tone signal is a version of the pilot tone signal that would be expected to be received if it propagated through a perfect communication channel, i.e., a communication channel having no multipath, a flat magnitude response and a zero phase response. Correlation is the most computationally straightforward of the three methods, but it is also the most sensitive to noise.
The BLUE method convolves the expected pilot tone signal with an estimate of the channel response, compares that result with the received pilot signal, and computes the difference between the two. The BLUE method then identifies the channel response that minimizes the mean square error between the expected pilot tone signal and the received pilot tone signal. This approach is more computationally intensive than correlation, in that a matrix, which incorporates the pilot data, must be inverted.
The MMSE method has elements in common with correlation and BLUE. Like BLUE, MMSE minimizes the mean square error between an expected pilot tone signal and a received pilot tone signal, and it also involves a matrix inversion. In addition, the noise power must be known at the receiver. At each channel estimation, the matrix must be reconstructed with the present noise power, requiring an additional complication. At high signal-to-noise ratios, MMSE resembles BLUE, but at low signal-to-noise ratios, MMSE resembles correlation.
Some channel characterization methods do not require as many pilot tones, or none at all. These methods are known as semi-blind and blind techniques, respectively.
Semi-blind techniques take advantage of feedback from a receiver. Turbo channel estimation and subspace-based methods are among the semi-blind approaches. Turbo channel estimation begins with a training set of pilot symbols, and then improves the channel estimation in succeeding iterations. Subspace-based methods invoke temporal correlations between consecutive data sets. All of these methods are effective at reducing the number of pilot tones that must be processed to achieve reliable channel characteristic estimation.
Blind techniques can eliminate use of pilot tones altogether. These methods fall into two categories: those based on Second Order Cyclostationary Statistics (SOCS) and those based on Higher Order Statistics (HOS). The SOCS technique requires the system to be stationary for a limited time period, and HOS requires signals at the receiver to be statistically independent. Therefore, SOCS may not be applicable to systems that are changing rapidly, and in general, HOS will not be readily applicable to MIMO (Multiple Input/Multiple Output) channels that have been Space-Time coded, because the coded signals are not statistically independent.
It would, therefore, be desirable to have a channel characterization method and system that can characterize a communication channel during a communication with minimum loss of channel bandwidth, and which can be used to identify channel characteristics for a communication channel that has dynamically changing characteristics.